Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~garcia/apj.537.l23.ps Äàòà èçìåíåíèÿ: Wed Feb 9 00:13:34 2011 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:18:39 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: dust disk

L23
The Astrophysical Journal, 537:L23--L26, 2000 July 1
# 2000. The American Astronomical Society. All rights reserved. Printed in U.S.A.
A FIRST LOOK AT THE NUCLEAR REGION OF M31 WITH CHANDRA
Michael R. Garcia, Stephen S. Murray, Francis A. Primini, William R. Forman,
Jeffrey E. McClintock, and Christine Jones
Harvard­Smithsonian Center for Astrophysics, MS­4, 60 Garden Street,
Cambridge, MA 02138; garcia@head­cfa.harvard.edu
Received 2000 March 13; accepted 2000 May 17; published 2000 June 23
ABSTRACT
We report on the first observation of the nuclear region of M31 with the Chandra X­Ray Observatory. The
nuclear source seen with the Einstein and ROSAT HRIs is resolved into five point sources. One of these sources
is within 1# of the M31 central supermassive black hole. As compared to the other point sources in M31, this
nuclear source has an unusual X­ray spectrum. Based on the spatial coincidence, we speculate that this source
is associated with the central black hole and note that the unusual spectrum is a challenge to some current
theories. A bright transient is detected #26# to the west of the nucleus and may be associated with a stellar mass
black hole.
Subject headings: black hole physics --- galaxies: individual (M31)
1. INTRODUCTION
As our nearest Milky Way analog, M31 offers us a chance to
study a galaxy like our own without the effects of severe ex­
tinction. For example, the nucleus of our Galaxy (Sagittarius A*)
is obscured by #30 mag of visual extinction (Morris & Serabyn
1996), while the nucleus of M31 likely suffers #2 mag of ex­
tinction (see § 3). In addition, the study of X­ray binaries in the
Galactic plane is hindered by extinction sometimes reaching
more than 30 mag, which can be compared to an average
mag for globular clusters in M31 (Barmby et al. 2000).
A # 0.7
V
Hubble Space Telescope (HST) observations resolved the
M31 nucleus into two components (P1 and P2) separated by
#0#.5 (Lauer et al. 1993). Ground­based measurements of the
rotational velocity of stars near the nucleus of M31 provide
strong evidence for a central dark, compact object of mass
M , , presumably a black hole (Kormendy & Bender
7
3.0 # 10
1999 and references therein). These observations support the
model of the double nucleus of M31 as a torus of stars orbiting
the nucleus in a slightly eccentric orbit (Tremaine 1995). Post­
COSTAR HST observations have shown that there is a group
of partially resolved UV­bright stars between P1 and P2 (Brown
et al. 1998) at the likely position of the central black hole
(Kormendy & Bender 1999).
The first identification of an X­ray source with the M31
nucleus came with Einstein observations, which found a source
within 2#.1 of the nucleus with ergs s #1
37
L = 9.6 # 10
X
(0.2--4.0 keV; van Speybroeck et al. 1979). Subsequent Einstein
observations showed the nucleus to be variable by factors of
#10 (Trinchieri & Fabbiano 1991) on timescales of 6 months.
ROSAT observations found a nuclear luminosity of L =
X
ergs s #1 (Primini, Forman, & Jones 1993).
37
2.1 # 10
Radio observations reveal a weak (#30 mJy) source at the
nucleus (Crane, Dickel, & Cowan 1992). The luminosity at
3.6 cm is # that of Sgr A*, a puzzle given that the M31
1
5
nucleus is #30 times more massive (Melia 1992). The corre­
lation between the radio and X­ray properties of low­luminosity
supermassive black holes (Yi & Boughn 1999) might be ex­
plained by an advection­dominated accretion flow (ADAF)
model, but M31 is an outlier in these correlations.
The point sources distributed throughout M31 are likely X­
ray binaries and supernova remnants similar to those in the
Galaxy. The fact that #40% of these sources are variable is
consistent with this hypothesis (Primini et al. 1993). Compar­
ison of Einstein and ROSAT images shows that #6% of the
sources are transient (Primini et al. 1993). A comparison of
Einstein and EXOSAT observations allowed discovery of two
additional transients (White & Peacock 1988).
The sensitivity and arcsecond spatial resolution of Chandra
(van Speybroeck et al. 1997; Weisskopf & O'Dell 1997) pro­
vide new insights into the X­ray properties of M31. A few of
those new insights, concerning the nucleus and a new transient,
are reported in this Letter.
2. OBSERVATIONS
Chandra was pointed at the nucleus of M31 for 17.5 ks on
1999 October 13. The Advanced CCD Imaging Spectrometer
(ACIS) imaging array (Nousek et al. 1998) was in the focal
plane and yielded a # image centered on the M31
#
16 # 16
nucleus. In this Letter, we concentrate on the observations of
the central #1# only. The ACIS­S3 chip counting rate was used
as an indicator of high background, and whenever it increased
above 1.5 counts s #1 we rejected the data. Consequently, we
obtained 8.8 ks of active observing time. Data were analyzed
with a combination of the Chandra X­Ray Center CIAO V1.1
(M. Elvis et al. 2000, in preparation), HEASARC XSPEC
V10.0 (Arnaud 1996), and software written by A. Vikhlinin
(Vikhlinin et al. 1998). Unless otherwise specified, all error
regions herein are 68% confidence bounds and include a 20%
uncertainty in the ACIS effective area below 0.27 keV. We note
that this calibration uncertainty is less than 50% of the statistical
uncertainties for the sources considered herein.
The ROSAT HRI imaged the central region of M31 six times
from 1990 to 1996, with exposure times ranging from 5 to
84.5 ks (see Primini et al. 1993; F. A. Primini et al. 2000, in
preparation). The image of the nuclear region from the last
observation is shown in Figure 1 (top).
3. DATA ANALYSIS
3.1. The Nucleus
The central object seen with the ROSAT HRI is clearly re­
solved into five sources by Chandra (Fig. 1, bottom). Based
on the Chandra aspect solution, which is currently limited by
systematics to #1# accuracy (see Aldcroft et al. 2000), we find

L24 FIRST LOOK AT M31 WITH CHANDRA Vol. 537
Fig. 1.---Top: Nuclear 64# of M31 as it appears in an 84.5 ks ROSAT HRI
observation in 1996 January. Bottom: Same as seen in an 8.8 ks Chandra
ACIS­I observation on 1999 October 13. The spatial resolution (FWHM) of
the ROSAT image is 7#, and that of the Chandra image is 0#.8; the pixel size
is 0#.5 in both cases.
TABLE 1
Positions of Nuclear Sources
Source
R.A.
(J2000)
Decl.
(J2000)
Nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 42 44.24 41 16 08.0
Transient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 42 42.03 41 16 08.8
CXOGMP J004247.0#411628 . . . . . . 00 42 47.04 41 16 28.8
CXOGMP J004244.2#411609 . . . . . . 00 42 44.23 41 16 09.2
CXOGMP J004444.2#411605 . . . . . . 00 44 44.23 41 16 05.8
CXOGMP J004243.9#411604 . . . . . . 00 42 43.97 41 16 04.7
CXOGMP J004243.7#411604 . . . . . . 00 42 43.72 41 16 04.4
Note.---Units of right ascension are hours, minutes, and seconds,
and units of declination are degrees, arcminutes, and arcseconds.
Absolute and relative positions are accurate to #1# and #0#.2,
respectively. Both are currently dominated by systematics. The
source naming convention follows that at http://asc.harvard.edu/
udocs/naming.html.
Fig. 2.---Hardness ratio for 79 sources with more than 20 total counts found
in the ACIS­S image of M31. The nuclear source, CXO J004244.2#411608,
has the third lowest hardness ratio and is indicated by the ``N.'' The nearby
transient is indicated by the ``T.'' The source #1# north of the nucleus is in the
first bin below 0.0, the source #2# to the south of the nucleus is in the bin
indicated by the ``T.''
that one of these five sources, CXO J004244.2#411608, is
within 1# of the position of the radio nucleus (Crane et al.
1992). As an independent check on the aspect, we computed
a plate solution for the X­ray image using the optical positions
of 10 X­ray--detected globular clusters from the Bologna cat­
alog (Battistini et al. 1987). This solution is limited by the
accuracy of the Bologna catalog and agrees (within the errors
of #0#.7 rms in right ascension and declination) with the Chan­
dra aspect. The positions of the nucleus and the other four
central sources based on the Chandra aspect are included in
Table 1.
We computed a wavelet deconvolution of the image and
found 121 point sources in the full field of view
#
16 # 16
(FOV) of ACIS­I (these sources will be discussed in a separate
paper). We then computed the hardness ratio of the 79 sources
with more than 20 counts; this ratio is defined as the counts
in the 1.0--7.0 keV band minus the counts in the 0.1--1.0 keV
band, divided the sum of the counts in both bands. This ratio
(Fig. 2) shows the nuclear source is one of three outliers with
extremely soft spectra.
When extracting the nuclear pulse height analysis (PHA) spec­
trum, we limited contamination from CXO J004244.2#411609,
which is only 1#.2 to the north, by excluding photons more than
0#.5 to the north of the nuclear source. We found a total of 100
counts in the 3 arcsec 2 region surrounding the nucleus. The re­
sulting PHA spectrum was fit with XSPEC, after first binning
the data such that each fitted bin had more than 10 counts.
Gehrels weighting was used for the fits (Gehrels 1986). The fits
were limited to the 0.3--1.8 keV region, since there were insuf­
ficient counts outside of this region.
Simple models (power law, blackbody, bremsstrahlung, with
interstellar absorption) provide acceptably good fits to the data.
The power­law fits give a slope and limit
#7
a = 5 N =
#2.4 H
cm #2 . In order to reduce the error range on the
#9 21
4 # 10
#3.5
fitted slope, we choose to limit the allowed range of absorption
to that found for the nearby transient (below), i.e., to N =
H
cm #2 . This then allows us to further restrict
21
(2.8 # 1.0) # 10
the slope (or temperature) of the spectrum to ,
a = 4.5 # 1.5
keV, or keV for power­law,
#0.06
kT = 0.15 kT = 0.43 # 0.17
#0.03
blackbody, or bremsstrahlung fits (respectively). A comparison
of the PHA spectra of the nuclear source, the transient, and
CXOGMP J004247.0#411628 can be seen in Figure 3.
The detected 0.3--7.0 keV flux of the nuclear source, assum­
ing the further restricted range of parameters for the power­
law model, is ergs cm #2 s #1 , corresponding to
#0.9 #14
5.8 # 10
#0.5
an observed luminosity of ergs s #1 at 784 kpc
#0.6 36
3.9 # 10
#0.3
(Stanek & Garnavich 1998). At the lowest N H and flattest a in
this range, approximately 60% of the 0.3--7.0 keV flux is ab­
sorbed by the interstellar medium, while at the highest N H and
steepest a, nearly 98% of the flux is absorbed. The corre­

No. 1, 2000 GARCIA ET AL. L25
Fig. 3.---PHA spectra of the nuclear source, the nearby transient, and
CXOGMP J004247.0#411628, one of the four nearby sources used as a probe
of the N H in the nuclear region.
sponding emitted luminosity ranges from to
37
1.2 # 10
ergs s #1 and has a nominal value at the best­fit
38
1.6 # 10
parameters of ergs s #1 . In order to test for the ex­
37
4.0 # 10
istence of a hard ( ) nuclear component, we note that three
a = 2
of the 100 detected photons have energies above 1.5 keV. This
does not constitute a significant detection above 1.5 keV. Mak­
ing the conservative assumption that these three photons are
due to an undetected hard component, we set a 90% upper
limit to the 1.5--7.0 keV luminosity of a hard component of
ergs s #1 .
35
7.6 # 10
In order to test our assumption that the N H measured for the
transient is appropriate to apply to the nucleus, we fit power­
law spectra to four other bright sources within 2# of the nucleus.
We selected sources approximately as bright as the transient
in order to accurately measure N H . In every case the 90%
confidence regions for N H overlap with the transient, therefore
indicating that our assumption is reasonable. We note that the
Galactic hydrogen column in the direction of M31 is only
cm #2 (Dickey & Lockman 1990), so our results
20
N # 7 # 10
H
are consistent with additional local absorption within M31 it­
self. If the gas/dust ratio in M31 is similar to that in the Galaxy,
the nuclear is equal to 1.5 # 0.6 (Predehl & Schmitt 1995).
A V
3.2. The Nearby Transient
We extracted 763 counts from a 1# radius circle at the po­
sition of CXO J004242.0#411608. Power­law, bremsstrah­
lung, and blackbody fits are all acceptable ( for 71
2
x /n ! 1.13
degrees of freedom), but the power­law fits produce the lowest
x 2 /n (#0.56). The best­fitting power­law number slope is
, with a best fit cm #2 . As­
21
1.5 # 0.3 N = (2.8 # 1.0) # 10
H
suming this model, the detected flux is #13
(7.4 # 0.7) # 10
ergs cm #2 s #1 , corresponding to an emitted luminosity of
ergs s #1 (0.3--7.0 keV). From the deepest
37
7.0 # 0.8 # 10
(and last) ROSAT HRI observation, and assuming the spectrum
above, we compute a 95% upper limit to the emitted luminosity
of ergs s #1 in the 0.3--7.0 keV band. Thus, the
36
3.0 # 10
transient brightened by at least a factor of #20.
4. DISCUSSION
4.1. The Nucleus
Based on the 1# positional coincidence of the Chandra nu­
clear source with the radio nucleus, we suggest that this X­ray
source may be associated with the central supermassive black
hole. However, we are currently unable to locate this source
with the accuracy necessary to locate it within P1, P2, or the
group of UV­bright stars in between, nor can we rule out an
association with some other object within #1# of the nucleus.
Previous authors have raised the possibility that the central
X­ray source may not be associated with the central black hole,
but is merely a chance coincidence (van Speybroeck et al. 1979;
Yi & Boughn 1999). The probability of a chance coincidence
depends on what search region one uses, and a posteriori it is
hard to know what the relevant search region is. If we use the
full ACIS FOV as the search region, then the chance of any
one of the 121 detected sources source being within 1# of the
nucleus is # . However, the surface density of sources
#4
4 # 10
increases toward the nucleus, so the chance probability may
be higher than this. The anomalously soft nature of the three
outliers in Figure 2 (one of which is the nucleus) suggests they
may be analogs to the supersoft X­ray sources found in our
galaxy, which are believed to be accreting white dwarfs (van
den Heuvel et al. 1992). The random probability of any one
of these three sources being within 1# of the nucleus is #10 #5 .
With only three anomalously soft sources detected, it is not
possible to determine if the density of these sources increases
toward the nucleus.
Several authors have previously noted the unusually low X­
ray and radio luminosity of the nucleus of M31 (Melia 1992;
Yi & Boughn 1999). We note that the X­ray luminosity we
find herein is substantially lower than that quoted in several
recent papers comparing X­ray and radio luminosities of low­
luminosity supermassive black holes (e.g., Franceschini, Ver­
cellone, & Fabian 1998; Yi & Boughn 1999). At this revised
luminosity, the M31 nucleus appears to be even more of an
outlier on the correlations between radio luminosity, X­ray lu­
minosity, and black hole mass found for low­luminosity su­
permassive black holes (Yi & Boughn 1999, Figs. 4 and 5).
We note that if the Chandra nuclear source is not associated
with the supermassive black hole, then the M31 nucleus must
be even fainter than we suggest.
While previous Einstein and ROSAT images are unable to
separate the nuclear source from the surrounding four sources,
the fluxes indicate that this composite source is highly variable.
In order to compare these fluxes to the Chandra flux, we assume
the nuclear power­law spectrum found above and use the count­
ing rates from the literature (van Speybroeck et al. 1979; Trin­
chieri & Fabbiano 1991; Primini et al. 1993) to calculate
0.2--4.0 keV detected fluxes. The uncertainty in the nuclear
spectrum allows up to 40% uncertainty in the conversion from
counting rate to flux. In order to make a fair comparison, Ta­

L26 FIRST LOOK AT M31 WITH CHANDRA Vol. 537
TABLE 2
M31 Nuclear X­Ray Flux
Date Observatory
Flux
(#10 #13 ergs cm #2 s #1 )
1979 Jan . . . . . . . Einstein 7.07 # 0.06
1979 Aug . . . . . . Einstein 0.60 # 0.18
1980 Jan . . . . . . . Einstein 3.50 # 0.64
1990 Jul . . . . . . . ROSAT 1.70 # 0.12
1999 Oct . . . . . . . Chandra a 1.43 # 0.15
a The Chandra flux is the sum of all five sources near the
nucleus.
ble 2 lists the summed flux from all five nuclear sources in the
Chandra image.
Strong variability of unresolved sources is often cited as
evidence for a small number of sources, simply because it is
more plausible for a single source to vary than for several
sources to vary coherently. If we apply this argument to the
M31 nucleus, it implies that one of these five sources is highly
variable. If we assume that it is the nucleus, then it is appro­
priate to assume that the average flux of the surrounding four
sources is constant and subtract this flux from the Einstein and
ROSAT measurements in order to determine the flux of the
nucleus alone. From the Chandra image, the flux from these
four sources is ergs cm #2 s #1 . Subtracting this,
#13
0.85 # 10
we see that the lowest Einstein flux measurement is consistent
with zero flux from the nucleus and indicates a factor of #40
variability.
As an aside, we note that the detection of Sgr A* with
Chandra (Garmire 1999) does not necessarily rule out a soft,
M31­like spectrum. Although the very high extinction to
Sgr A* would reduce the count rate for our best­fit spectrum
by #60 times, nevertheless, this would be more than offset by
the #100 times smaller distance.
Standard ADAF models are not able to explain the ratio of
X­ray to radio luminosity of the nucleus (Yi & Boughn 1999).
However, models including winds (Di Matteo et al. 2000) and/
or convective flows (Narayan, Igumenshchev, & Abramowicz
1999) may be able to explain this ratio. These models generally
predict hard spectra in the X­ray region, so they may not be
able to explain the extremely soft spectrum reported herein
(E. Quataert 2000, private communication). A class of models
which may be able to explain the soft spectra and which nat­
urally predict strong and rapid variability are those in which
the emission is due to a jet, rather than accretion (i.e., H. Falcke
& S. Markoff 2000, in preparation; Falcke 1996).
4.2. The Nearby Transient
The location of the transient in the bulge of M31 suggests
it is an X­ray nova: stars in the inner bulge of M31 are likely
old, disk/bulge population stars typical of those in X­ray novae
(White, Nagase, & Parmar 1995; Tanaka & Lewin 1995), rather
than the young, Be stars typically found in star­forming regions
and in X­ray transient Be­star pulsar systems. X­ray novae
typically consist of a black hole accreting from a low­mass
companion. We note that in either case the optical magnitude
of the transient in outburst is likely to be , making the
V # 22
object visible with HST. An X­ray nova would be expected to
show a large variation in V from quiescence to outburst, while
a Be­star pulsar would show a more moderate variation. HST
observations are underway in an attempt to clarify the nature
of this transient.
We thank the CXC team for help with ACIS data analysis
and reductions and the entire Chandra team for building a
superb observatory. Antonella Fruscione and Norbert Shultz
deserve extra thanks for their help with the analysis software
and calibrations, respectively. We thank the referee, Nicholas
White, for several insightful and helpful suggestions. This work
was supported in part by NASA contract NAS8­39073.
REFERENCES
Aldcroft, T., Karovska, M., Cresitello­Dittmar, M., Cameron, R., &Markevitch,
M. 2000, Proc. SPIE, 4012, 74
Arnaud, K. A. 1996, in ASP Conf. Ser. 101, Astronomical Data Analysis
Software and Systems V, ed. G. H. Jacoby & H. Barnes (San Francisco:
ASP), 17
Barmby, P., Huchra, J. P., Brodie, J. P., Forbes, D. A., Schroder, L. L., &
Grillmair, C. J. 2000, AJ, 119, 727
Battistini, P., et al. 1987, A&AS, 67, 447
Brown, T. M., Ferguson, H. C., Stanford, S. A., & Deharveng, J.­M. 1998,
ApJ, 504, 113
Crane, P. C., Dickel, J. R., & Cowan, J. J. 1992, ApJ, 390, L9
Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215
Di Matteo, T., Quataert, E., Allen, S. W., Narayan, R., & Fabian, A. C. 2000,
MNRAS, submitted (astro­ph/9905053)
Falcke, H. 1996, ApJ, 464, L67
Franceschini, A., Vercellone, S., & Fabian, A. C. 1998, MNRAS, 297, 817
Garmire, G. P. 1999, BAAS, 195, 9603
Gehrels, N. 1986, ApJ, 303, 336
Kormendy, J., & Bender, R. 1999, ApJ, 522, 772
Lauer, T. R., et al. 1993, AJ, 106, 1436
Melia, F. 1992, ApJ, 398, L95
Morris, M., & Serabyn, E. 1996, ARA&A, 34, 645
Narayan, R., Igumenshchev, I. V., & Abramowicz, M. A. 1999, preprint (astro­
ph/9912449)
Nousek, J. A., et al. 1998, Proc. SPIE, 3444, 225
Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889
Primini, F. A., Forman, W., & Jones, C. 1993, ApJ, 410, 615
Stanek, K. Z., & Garnavich, P. M. 1998, ApJ, 503, L131
Tanaka, Y., & Lewin, W. H. G. 1995, in X­Ray Binaries, ed. W. H. G. Lewin,
J. van Paradijs, & E. P. J. van den Heuvel (Cambridge: Cambridge Univ.
Press), 126
Tremaine, S. 1995, AJ, 110, 628
Trinchieri, G., & Fabbiano, G. 1991, ApJ, 382, 82
van den Heuvel, E. P. J., Bhattachary, D., Nomoto, K., & Rappaport, S. A.
1992, A&A, 262, 97
van Speybroeck, L., Epstein, A., Forman, W., Giacconi, R., Jones, C., Liller,
W., & Smarr, L. 1979, ApJ, 234, L45
van Speybroeck, L., Jerius, D., Edgar, R. J., Gaetz, T. J., Zhao, P., & Reid,
P. B. 1997, Proc. SPIE, 3113, 98
Vikhlinin, A., McNamara, B. R., Forman, W., Jones, C., Quintana, H., &
Hornstrup, A. 1998, ApJ, 502, 558
Weisskopf, M. C., & O'Doell, S. L. 1997, Proc. SPIE, 3113, 2
White, N. E., Nagase, F., & Parmar, A. N. 1995, in X­ray Binaries, ed.
W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel (Cambridge:
Cambridge Univ. Press), 58
White, N. E., & Peacock, A. 1998, Mem. Soc. Astron. Italiana, 59, 7
Yi, I., & Boughn, S. P. 1999, ApJ, 515, 576